Cross-Contrastive Clustering for Multimodal Attributed Graphs with Dual Graph Filtering
This work solves the challenge of clustering complex multimodal graph data for applications like social community detection and medical analytics, representing an incremental advance over prior multi-view clustering methods.
The paper tackled the problem of clustering multimodal attributed graphs (MMAGs) by addressing low modality-wise correlation and feature-wise noise from pre-trained models, which existing methods overlook, and proposed a Dual Graph Filtering scheme with tri-cross contrastive training, achieving significant improvements in clustering quality on eight benchmark datasets.
Multimodal Attributed Graphs (MMAGs) are an expressive data model for representing the complex interconnections among entities that associate attributes from multiple data modalities (text, images, etc.). Clustering over such data finds numerous practical applications in real scenarios, including social community detection, medical data analytics, etc. However, as revealed by our empirical studies, existing multi-view clustering solutions largely rely on the high correlation between attributes across various views and overlook the unique characteristics (e.g., low modality-wise correlation and intense feature-wise noise) of multimodal attributes output by large pre-trained language and vision models in MMAGs, leading to suboptimal clustering performance. Inspired by foregoing empirical observations and our theoretical analyses with graph signal processing, we propose the Dual Graph Filtering (DGF) scheme, which innovatively incorporates a feature-wise denoising component into node representation learning, thereby effectively overcoming the limitations of traditional graph filters adopted in the extant multi-view graph clustering approaches. On top of that, DGF includes a tri-cross contrastive training strategy that employs instance-level contrastive learning across modalities, neighborhoods, and communities for learning robust and discriminative node representations. Our comprehensive experiments on eight benchmark MMAG datasets exhibit that DGF is able to outperform a wide range of state-of-the-art baselines consistently and significantly in terms of clustering quality measured against ground-truth labels.